Implementing some of java.lang.Character methods

teavm-classlib/src/main/java/org/teavm/classlib/java/lang/ClassNativeGenerator.java

@@ -83,6 +83,9 @@ public class ClassNativeGenerator implements Generator, Injector, DependencyPlug
             case "voidClass":
                 context.getWriter().append("$rt_cls($rt_voidcls())");
                 break;
+            case "charClass":
+                context.getWriter().append("$rt_cls($rt_charcls())");
+                break;
             case "wrap":
                 context.writeExpr(context.getArgument(0));
                 break;
@@ -172,6 +175,7 @@ public class ClassNativeGenerator implements Generator, Injector, DependencyPlug
         switch (graph.getReference().getName()) {
             case "booleanClass":
             case "intClass":
+            case "charClass":
             case "wrap":
             case "getSuperclass":
             case "getComponentType0":

teavm-classlib/src/main/java/org/teavm/classlib/java/lang/TCharacter.java

@@ -28,8 +28,73 @@ import org.teavm.javascript.ni.Rename;
 public class TCharacter extends TObject {
     public static final int MIN_RADIX = 2;
     public static final int MAX_RADIX = 36;
+    public static final char MIN_VALUE = '\0';
+    public static final char MAX_VALUE = '\uFFFF';
+    public static final TClass<TCharacter> TYPE = TClass.charClass();
+    public static final byte UNASSIGNED = 0;
+    public static final byte UPPERCASE_LETTER = 1;
+    public static final byte LOWERCASE_LETTER = 2;
+    public static final byte TITLECASE_LETTER = 3;
+    public static final byte MODIFIER_LETTER = 4;
+    public static final byte OTHER_LETTER = 5;
+    public static final byte NON_SPACING_MARK = 6;
+    public static final byte ENCLOSING_MARK = 7;
+    public static final byte COMBINING_SPACING_MARK = 8;
+    public static final byte DECIMAL_DIGIT_NUMBER = 9;
+    public static final byte LETTER_NUMBER = 10;
+    public static final byte OTHER_NUMBER = 11;
+    public static final byte SPACE_SEPARATOR = 12;
+    public static final byte LINE_SEPARATOR = 13;
+    public static final byte PARAGRAPH_SEPARATOR = 14;
+    public static final byte CONTROL = 15;
+    public static final byte FORMAT = 16;
+    public static final byte PRIVATE_USE = 17;
+    public static final byte SURROGATE = 19;
+    public static final byte DASH_PUNCTUATION = 20;
+    public static final byte START_PUNCTUATION = 21;
+    public static final byte END_PUNCTUATION = 22;
+    public static final byte CONNECTOR_PUNCTUATION = 23;
+    public static final byte OTHER_PUNCTUATION = 24;
+    public static final byte MATH_SYMBOL = 25;
+    public static final byte CURRENCY_SYMBOL = 26;
+    public static final byte MODIFIER_SYMBOL = 27;
+    public static final byte OTHER_SYMBOL = 28;
+    public static final byte INITIAL_QUOTE_PUNCTUATION = 29;
+    public static final byte FINAL_QUOTE_PUNCTUATION = 30;
+    public static final byte DIRECTIONALITY_UNDEFINED = -1;
+    public static final byte DIRECTIONALITY_LEFT_TO_RIGHT = 0;
+    public static final byte DIRECTIONALITY_RIGHT_TO_LEFT = 1;
+    public static final byte DIRECTIONALITY_RIGHT_TO_LEFT_ARABIC = 2;
+    public static final byte DIRECTIONALITY_EUROPEAN_NUMBER = 3;
+    public static final byte DIRECTIONALITY_EUROPEAN_NUMBER_SEPARATOR = 4;
+    public static final byte DIRECTIONALITY_EUROPEAN_NUMBER_TERMINATOR = 5;
+    public static final byte DIRECTIONALITY_ARABIC_NUMBER = 6;
+    public static final byte DIRECTIONALITY_COMMON_NUMBER_SEPARATOR = 7;
+    public static final byte DIRECTIONALITY_NONSPACING_MARK = 8;
+    public static final byte DIRECTIONALITY_BOUNDARY_NEUTRAL = 9;
+    public static final byte DIRECTIONALITY_PARAGRAPH_SEPARATOR = 10;
+    public static final byte DIRECTIONALITY_SEGMENT_SEPARATOR = 11;
+    public static final byte DIRECTIONALITY_WHITESPACE = 12;
+    public static final byte DIRECTIONALITY_OTHER_NEUTRALS = 13;
+    public static final byte DIRECTIONALITY_LEFT_TO_RIGHT_EMBEDDING = 14;
+    public static final byte DIRECTIONALITY_LEFT_TO_RIGHT_OVERRIDE = 15;
+    public static final byte DIRECTIONALITY_RIGHT_TO_LEFT_EMBEDDING = 16;
+    public static final byte DIRECTIONALITY_RIGHT_TO_LEFT_OVERRIDE = 17;
+    public static final byte DIRECTIONALITY_POP_DIRECTIONAL_FORMAT = 18;
+    public static final char MIN_HIGH_SURROGATE = '\uD800';
+    public static final char MAX_HIGH_SURROGATE = '\uDBFF';
+    public static final char MIN_LOW_SURROGATE  = '\uDC00';
+    public static final char MAX_LOW_SURROGATE  = '\uDFFF';
+    public static final char MIN_SURROGATE = MIN_HIGH_SURROGATE;
+    public static final char MAX_SURROGATE = MAX_LOW_SURROGATE;
+    public static final int MIN_SUPPLEMENTARY_CODE_POINT = 0x010000;
+    public static final int MIN_CODE_POINT = 0x000000;
+    public static final int MAX_CODE_POINT = 0X10FFFF;
+    public static final int SIZE = 16;
+    static final int ERROR = 0xFFFFFFFF;
     private static int[] digitMapping;
     private char value;
+    private static TCharacter[] characterCache = new TCharacter[128];
 
     public TCharacter(char value) {
         this.value = value;
@@ -40,13 +105,21 @@ public class TCharacter extends TObject {
     }
 
     public static TCharacter valueOf(char value) {
+        if (value < characterCache.length) {
+            TCharacter result = characterCache[value];
+            if (result == null) {
+                result = new TCharacter(value);
+                characterCache[value] = result;
+            }
+            return result;
+        }
         return new TCharacter(value);
     }
 
     @Override
     @Rename("toString")
     public TString toString0() {
-        return new TString(new char[] { value });
+        return toString(value);
     }
 
     @Override
@@ -57,6 +130,100 @@ public class TCharacter extends TObject {
         return other instanceof TCharacter && ((TCharacter)other).value == value;
     }
 
+    @Override
+    public int hashCode() {
+        return value;
+    }
+
+    public static TString toString(char c) {
+        return new TString(new char[] { c });
+    }
+
+    public static boolean isValidCodePoint(int codePoint) {
+        return codePoint >= 0 && codePoint <= MAX_CODE_POINT;
+    }
+
+    public static boolean isBmpCodePoint(int codePoint) {
+        return codePoint > 0 && codePoint <= MAX_VALUE;
+    }
+
+    public static boolean isSupplementaryCodePoint(int codePoint) {
+        return codePoint >= MIN_SUPPLEMENTARY_CODE_POINT && codePoint <= MAX_CODE_POINT;
+    }
+
+    public static boolean isHighSurrogate(char ch) {
+        return UTF16Helper.isHighSurrogate(ch);
+    }
+
+    public static boolean isLowSurrogate(char ch) {
+        return UTF16Helper.isLowSurrogate(ch);
+    }
+
+    public static boolean isSurrogate(char ch) {
+        return isHighSurrogate(ch) || isLowSurrogate(ch);
+    }
+
+    public static boolean isSurrogatePair(char high, char low) {
+        return isHighSurrogate(high) && isLowSurrogate(low);
+    }
+
+    public static int charCount(int codePoint) {
+        return isSupplementaryCodePoint(codePoint) ? 2 : 1;
+    }
+
+    public static int toCodePoint(char high, char low) {
+        return UTF16Helper.buildCodePoint(high, low);
+    }
+
+    public static int codePointAt(TCharSequence seq, int index) {
+        if (index >= seq.length() - 1 || !isHighSurrogate(seq.charAt(index)) ||
+                !isLowSurrogate(seq.charAt(index + 1))) {
+            return seq.charAt(index);
+        } else {
+            return toCodePoint(seq.charAt(index), seq.charAt(index + 1));
+        }
+    }
+
+    public static int codePointAt(char[] a, int index) {
+        return codePointAt(a, index, a.length);
+    }
+
+    public static int codePointAt(char[] a, int index, int limit) {
+        if (index >= limit - 1 || !isHighSurrogate(a[index]) || !isLowSurrogate(a[index + 1])) {
+            return a[index];
+        } else {
+            return toCodePoint(a[index], a[index + 1]);
+        }
+    }
+
+    public static int codePointBefore(TCharSequence seq, int index) {
+        if (index == 0 || !isLowSurrogate(seq.charAt(index)) || !isHighSurrogate(seq.charAt(index - 1))) {
+            return seq.charAt(index);
+        } else {
+            return toCodePoint(seq.charAt(index - 1), seq.charAt(index));
+        }
+    }
+
+    public static int codePointBefore(char[] a, int index) {
+        return codePointBefore(a, index, 0);
+    }
+
+    public static int codePointBefore(char[] a, int index, int start) {
+        if (index <= start || !isLowSurrogate(a[index]) || !isHighSurrogate(a[index - 1])) {
+            return a[index];
+        } else {
+            return toCodePoint(a[index - 1], a[index]);
+        }
+    }
+
+    public static char highSurrogate(int codePoint) {
+        return UTF16Helper.highSurrogate(codePoint);
+    }
+
+    public static char lowSurrogate(int codePoint) {
+        return UTF16Helper.lowSurrogate(codePoint);
+    }
+
     @GeneratedBy(CharacterNativeGenerator.class)
     public static native char toLowerCase(char ch);
 
@@ -138,6 +305,54 @@ public class TCharacter extends TObject {
         }
     }
 
+    public static int codePointCount(TCharSequence seq, int beginIndex, int endIndex) {
+        int count = endIndex;
+        --endIndex;
+        for (int i = beginIndex; i < endIndex; ++i) {
+            if (UTF16Helper.isHighSurrogate(seq.charAt(i)) && UTF16Helper.isLowSurrogate(seq.charAt(i + 1))) {
+                --count;
+                ++i;
+            }
+        }
+        return count;
+    }
+
+    public static int codePointCount(char[] a, int offset, int count) {
+        int r = count;
+        --count;
+        for (int i = 0; i < count; ++i) {
+            if (UTF16Helper.isHighSurrogate(a[offset]) && UTF16Helper.isLowSurrogate(a[offset + i + 1])) {
+                --r;
+                ++i;
+            }
+        }
+        return r;
+    }
+
+    public static int offsetByCodePoints(TCharSequence seq, int index, int codePointOffset) {
+        for (int i = 0; i < codePointOffset; ++i) {
+            if (index < seq.length() - 1 && UTF16Helper.isHighSurrogate(seq.charAt(index)) &&
+                    UTF16Helper.isLowSurrogate(seq.charAt(index + 1))) {
+                index += 2;
+            } else {
+                index++;
+            }
+        }
+        return index;
+    }
+
+    public static int offsetByCodePoints(char[] a, int start, int count, int index, int codePointOffset) {
+        for (int i = 0; i < codePointOffset; ++i) {
+            if (index < count - 1 && UTF16Helper.isHighSurrogate(a[index + start]) &&
+                    UTF16Helper.isLowSurrogate(a[index + start + 1])) {
+                index += 2;
+            } else {
+                index++;
+            }
+        }
+        return index;
+    }
+
     public static boolean isISOControl(char ch) {
         return isISOControl((int)ch);
     }

teavm-classlib/src/main/java/org/teavm/classlib/java/lang/TClass.java

@@ -73,6 +73,10 @@ public class TClass<T> extends TObject {
     @PluggableDependency(ClassNativeGenerator.class)
     static native TClass<TBoolean> booleanClass();
 
+    @InjectedBy(ClassNativeGenerator.class)
+    @PluggableDependency(ClassNativeGenerator.class)
+    static native TClass<TCharacter> charClass();
+
     @InjectedBy(ClassNativeGenerator.class)
     @PluggableDependency(ClassNativeGenerator.class)
     static native TClass<TInteger> intClass();

teavm-classlib/src/main/java/org/teavm/classlib/java/lang/TString.java

@@ -118,42 +118,19 @@ public class TString extends TObject implements TSerializable, TComparable<TStri
     }
 
     public int codePointAt(int index) {
-        if (index == characters.length - 1 || !UTF16Helper.isHighSurrogate(characters[index]) ||
+        return TCharacter.codePointAt(this, index);
-                !UTF16Helper.isLowSurrogate(characters[index + 1])) {
-            return characters[index];
-        }
-        return UTF16Helper.buildCodePoint(characters[index], characters[index + 1]);
     }
 
     public int codePointBefore(int index) {
-        if (index == 1 || !UTF16Helper.isHighSurrogate(characters[index]) ||
+        return TCharacter.codePointBefore(this, index);
-                UTF16Helper.isLowSurrogate(characters[index - 1])) {
-            return characters[index - 1];
-        }
-        return UTF16Helper.buildCodePoint(characters[index - 1], characters[index]);
     }
 
     public int codePointCount(int beginIndex, int endIndex) {
-        int count = endIndex;
+        return TCharacter.codePointCount(this, beginIndex, endIndex);
-        --endIndex;
-        for (int i = beginIndex; i < endIndex; ++i) {
-            if (UTF16Helper.isHighSurrogate(characters[i]) && UTF16Helper.isLowSurrogate(characters[i + 1])) {
-                --count;
-            }
-        }
-        return count;
     }
 
     public int offsetByCodePoints(int index, int codePointOffset) {
-        for (int i = 0; i < codePointOffset; ++i) {
+        return TCharacter.offsetByCodePoints(this, index, codePointOffset);
-            if (index < characters.length - 1 && UTF16Helper.isHighSurrogate(characters[index]) &&
-                    UTF16Helper.isLowSurrogate(characters[index + 1])) {
-                index += 2;
-            } else {
-                index++;
-            }
-        }
-        return index;
     }
 
     @Override